Game development is an iterative process, and while this does grant you a good amount of freedom to experiment with new ideas. How much time you devote to each iteration is extremely important since each one inevitably costs you development time, so separating ideas that work from ones that don't quickly becomes paramount. Obviously you won't always know that what you're attempting is in fact a poor decision until you test it out, and there's no shame in that; sometimes things simply aren't as fun as you imagined they would be. More importantly just because the concept is bad in the current game you're working on doesn't always mean its a bad idea in general so you can keep it around until you find a better place for it later. Where there is a serious issue though, is when you've convinced yourself certain aspects of the game must work no matter what. What is important isn't whether or not you personally like the concept, but rather: 1. Whether the concept works in the first place. 2. Does it play well with the game thematically and in tone. 3. Finally whether other people agree that the mechanic is fun within the scope of the game. The final one is one of the parts that can get you into trouble, because to often people will hear feedback along the lines of "Well this thing is fun, but it doesn't feel like it fits." From that feedback the instinct might be to make sure it gets into the game period because you were told its "fun" but even then you may only end up muddying the game.
Unless you've been under a rock for the last year, Overwatch is Blizzard's next big project. Essentially Overwatch is a marriage between the team based shoot and the MOBA genre of games, though many people would probably cite games like Team Fortress 2 and Heros of the Storm to be more specific. It's a game that I'm really excited for and will be out in the very near future. Each and every character has been given a unique set of skills that are all based around a central theme to each character.